Three Big Data Myths and Eight Big Data Practices

1. Forget Big Data! If big data has become the norm for everyone to use data, why do you need to talk about him in particular?98 when the Internet is a buzzword, and now there are still people who will say that he is a buzzword? Now there are many e-commerce companies called traditional e-commerce ah, how sad ah, some people still think that e-commerce is something new when there has been the so-called traditional Internet companies.

? 2. data is only part of the innovative decision-making, he is only a new tool, and do not have to think of him as so omnipotent so God. Not all the problems are data problems, not all the problems are big data problems, you think of him as a simple tool to use, the knife with a knife, the gun with a gun, some places will be more suitable for the use of data, do not have to be too deified him, too many outsiders to speak of him as a god, but we dare not say that the industry is too mythical, because we know that can not be realized. We know that we can't fulfill it. 3. Do not data for data. We used to do a B2B website, the unit price keeps falling, we use a lot of data to solve the problem, but there is no improvement, one morning I think it is not right, I said we do not look at the data, I told the engineers, you ask the customer into the site when he asked: "You are to help themselves to buy things? As a result, more than 50% of the people said yes, you know I spent half a year to find the answer, this is simply for the sake of data for data ah, so if you are very worried about big data today, you might as well worry about the future there are a lot of people who will be for the sake of data for data ? Eight big data practical secret techniques ? 1. Do not talk about big data, let's say we use the data in the end when we know what the data behind the data? If my judgment is right, what data do you want to use to prove that I am right? For example, if the weather bureau says today's temperature is 12 degrees, what is the probability of being correct in predicting 12 degrees in this environment in the past? What is the probability of being wrong? This is the data of the data, I want to use a data before, I must ask, this data can not be trusted ah, how reliable is the reliability of the data, there is no reliability of the data, you first use, you are blindly use, so the data of the data is a level, decide whether it is reliable. So the data is a level that determines whether it is reliable or not. 2. Small and medium-sized enterprises to quantify themselves with data before talking about big data. How to use good data to quantify themselves? Use data to understand themselves, quantify themselves, I think on this basis to think about what we have what can be used with big data? That would be more effective. Small businesses should try to use data to quantify the decision, not big data, no big thing, is to use data to make decisions, in fact, the company itself has a lot of data problems within the company, such as most of the company's customer service data has never been connected with the company's main data, because many companies' customer service centers are in the outside, so he can not get the data, he does not know the consumer's reaction, this data and can not be associated with your operational data, so the whole data in the data to do correlation, so the whole data in the data to do correlation. data to do correlation, so the whole data in a small and medium-sized enterprises is also fragmented, you do not use this place in the case, you actually say you want to use big data, in fact, is a little difficult to understand. The data case will fail in many cases. 3. Many data cases fail because data collection is collection, but there is no way to integrate the data collection with the original data decision-making. This is not only an offline problem, but also an online problem. You can ask the person who manages the homepage now, how much of his homepage is designed according to the data, and you might as well ask them if they revamp the homepage, how do they evaluate the success of the revamped homepage? What data to decide? 4. What is the refresh frequency of the data? This value is very critical. Refresh fast is not necessarily better than slow, some places to brush a little slower. Some things are too sensitive, you refresh the data is not necessarily correct, for example, you want to buy a twenty-year insurance, is something very long-term, or you want to make a major investment, at this time you should go to look at the historical stability of the data, if today you have just watched a play from the cinema, you have just watched the end of want to eat hot spicy hot pot, this second, you do not need to guess his history of the character, you should go to guess what his next scene will be, the next scene. This time the location of the data is very critical. 5. real data innovation has not yet appeared, now most of the enterprises can not string data, algorithms and application of innovation, no one understands both business and data, to catch this point of opportunity is very few, I am, but I am only an expert in the field of e-commerce and retail. 6. Innovation in data comes from two things: one, splitting data that shouldn't be split anymore; and two, putting together two pieces of data that shouldn't be put together. This will generate a lot of power, for example, gender is either male or female, these two things should never be able to be split again, but in the data we can say that 30% of this person's attitudes are very girly, and 70% of them are very boyish, and his attitudes have boyish attitudes in them. Some of the data is already loess, but when you cut it open, you realize that it's not loess, it can be split again, and the destructive power or the creativity of the data that's been split out at this point is huge, and you've never thought of that, and that's so critical for recommendation engines. The business world is very competitive, when two rival companies are allied, such as adjusting premiums based on driving data, it is an innovative case of data combination. 7. I see big data projects are more disappointing, many big data projects are still in the laboratory, when these things to the enterprise will not work, the enterprise needs to be accurate, there are many problems are to be divided into scenarios. 8. data analysts to quantify their own quantification, which is very important to our industry. Our entire industry hates what you know? You find a person to accurately calculate a thing, it is not difficult, but six months are accurate, it is difficult, a long time, it is not accurate. Over time, the whole model is built out of historical data, and as the historical data becomes less and less important, the model becomes less accurate, and that's when you have to improve your algorithms.

The above is what I have shared with you about the three big data myths and the eight big data practical secret techniques, more information can be concerned about the Global Green Ivy to share more dry goods